I'm reading the Robinson and Smyth's 2008 paper: Small-sample estimation of negative binomial dispersion, with applications to SAGE data
Wondering if anyone can clarify the final test being used to determine differential expression for me? The variant of fishers exact.
As I'm reading it, from section 5.2 (and from the help(exactTest()), it seems that the adjusted counts (Za and Zb) for the two conditions are added together to form another NB distribution and then it tests where in the summed distribution, the counts for Za are? That seems ... overly simple.
Is that right? Am I even looking in the right place for an explanation of how this works? Robinson et. al's 2010 edger note suggest this paper. And I can't see it clearly explained in the Zhou, Lindsay, Robinson 2014 paper either.
Cheers
Ben.
Wondering if anyone can clarify the final test being used to determine differential expression for me? The variant of fishers exact.
As I'm reading it, from section 5.2 (and from the help(exactTest()), it seems that the adjusted counts (Za and Zb) for the two conditions are added together to form another NB distribution and then it tests where in the summed distribution, the counts for Za are? That seems ... overly simple.
Is that right? Am I even looking in the right place for an explanation of how this works? Robinson et. al's 2010 edger note suggest this paper. And I can't see it clearly explained in the Zhou, Lindsay, Robinson 2014 paper either.
Cheers
Ben.